Steven M. Searcy

Continuous Phase Modulation for Fiber-Optic Links

Advances in photonics, silicon electronics and digital signal processing (DSP) have converged to enable highly efficient transmission across fiber optic channels. Single wavelength data rates of 112 Gb/s are sought for wide deployment based on QPSK transmission, coherent detection, and digital demodulation. Here we examine continuous phase modulation (CPM) as a means to enhance performance and reach of coherent optical links. We quantify the robustness of the constant amplitude CPM format to spectral filtering and nonlinearities in comparison to QPSK. The challenges of generating and receiving the CPM waveform are considered and a novel CPM transmitter architecture is proposed.

Scaling 112 Gb/s Optical Networks With the Nonlinear Threshold Metric

A nonlinear penalty threshold that contains the total nonlinear phase shift is examined as a criterion for scaling fiber optic links. We investigate 112 Gb/s PDM-QPSK hybrid optical networks consisting of either TrueWave (G.655) or AllWave (G.652) fiber. We establish experimental and simulation environments with robust absolute matching in both linear and nonlinear regimes. We identify both XPM and SPM penalties in 0% and 100% inline dispersion compensation schemes for the two different fiber types. Both experimental and simulation results reveal that 0% inline-compensated links operate with larger nonlinear threshold for increasing span count and readily yield simple design rules.

Avoiding fiber nonlinearities by choice of modulation format

Nonlinear refraction in fiber optic links is a capacity limiting mechanism, whereby the phase of each propagating signal is modulated by intensity variations of signals in nearby channels. The transition to coherent detection enables a wide variety of modulation formats to be considered. Indeed, the choice of modulation format plays a primary role in determining the degree of amplitude variation in the channel as well as the robustness to the phase noise impairment that nonlinearities induce. On one hand, constant envelope formats (or nearly-constant) avoid fluctuations in the signal and produce lower nonlinearity-based impairments. Alternatively, star-QAM modulation formats enhance the receivers robustness to phase noise. Using simulated and experimental results we demonstrate the effectiveness of each format in avoiding fiber nonlinearity effects for both standard fiber (17ps/nm-km) and NZDF (5 ps/nm-km). We show sensitivity of several formats to nonlinear phase modulation from adjacent channels. We show the interaction between dispersion and constant envelope formats that guides the applications in which constant envelope formats, such as continuous phase modulation (CPM) provide gain over non-constant formats, such as QPSK. Consideration is made to scaling to 100 Gb/s and beyond in practical implementations.

Offset QPSK receiver implementation in 112 Gb/s coherent optical networks

We demonstrate Offset-QPSK transmitter and receiver architectures for 112 Gb/s coherent optical networks, highlighting similarities and differences to QPSK implementations. Experimentally we demonstrate that O-QPSK exhibits an enhanced immunity to nonlinearities when transmitted over TrueWave® fiber links.

System design tradeoffs for XPM mitigation in hybrid 100G-10G networks

The effectiveness of XPM reduction in 100G-10G networks employing fiber Bragg grating and fiber-based DCMs is determined experimentally as a function of key system parameters including dispersion map, spectral guard bands, and DSP parameters.

Scaling 100G QPSK links for reliable network development

Nonlinearities are a performance limitation in coherent optical links, and efforts have been made to understand the tradeoffs between launch power and the penalties related to nonlinearities. Using both simulation and experimental results from our 100G testbed we investigate the use of a nonlinear phase criterion that quantifies the total nonlinear phase accumulation within a 112 Gb/s PDM-QPSK link. We examine the nonlinear effects of self-phase (SPM) and cross-phase modulation (XPM) on a 112 Gb/s PM-QPSK channel propagating between four 10 Gb/s OOK aggressor channels on a 50 GHz grid and quantify the launch power and span count scaling behavior. In order to assess the applicability of a nonlinear phase criterion on real-world links, we determine the launch power that yields a 1.5 dB OSNR penalty at a BER of 10-3 for each configuration. This launch power then allows the identification of a Nonlinear Threshold Power (number of spans times launch power) that fully incorporates the increasing nonlinear penalties with further transmission distance. This metric allows for the determination of a set of engineering rules for deployment of 100 Gb/s PDM-QPSK in linear links with arbitrary number of spans and span distances. We find that this nonlinear threshold is constant in dispersion-compensated links. These experimental results are validated with simulations.

Offset QPSK for 112 Gb/s coherent optical links

Offset QPSK is a linear modulation format exhibiting nearly constant envelop and hence may avoid nonlinear refraction effects and thus enable higher launch power and a commensurate increase in reach. We experimentally demonstrate >2dB advantage for OQPSK compared to QPSK for multi span coherent optical links using TrueWave® RS fiber.

Scaling 112 Gb/s PDM-QPSK hybrid optical networks

We experimentally and numerically determined the nonlinear transmission impairments for a hybrid PDM-QPSK/OOK network. Results are quantified via a nonlinear phase metric and can be used as design rules for scaling hybrid and other links.

Asynchronously sampled blind source separation for coherent optical links

Polarization multiplexing is an integral technique for generating spectrally efficient 100 Gb/s and higher optical links. Post coherent detection DSP-based polarization demultiplexing of QPSK links is commonly performed after timing recovery. We propose and demonstrate a method of asynchronous blind source separation using the constant modulus algorithm (CMA) on the asynchronously sampled signal to initially separate energy from arbitrarily aligned polarization states. This method lends well to implementation as it allows for an open-loop sampling frequency for analog-to-digital conversion at less than twice the symbol rate. We show that the performance of subsequent receiver functions is enhanced by the initial pol demux operation. CMA singularity behavior is avoided through tap settling constraints. The method is applicable to QPSK transmissions and many other modulation formats as well, including general QAM signals, offset-QPSK, and CPM, or a combination thereof. We present the architecture and its performance under several different formats and link conditions. Comparisons of complexity and performance are drawn between the proposed architecture and conventional receivers.

Quaternary modulation formats for 100-Gbps optical links

The demand for 100 Gb/s optical links is rapidly spreading across all levels of the optical networking infrastructure. Many of the first deployments will be in the local area network (LAN) and metro-core and regional network environments. To address needs in LAN, the upcoming IEEE standard (IEEE P802.3ba) seeks 100 Gb/s over distances up to 40km. Furthermore metro-core/regional dense wavelength division multiplexing (DWDM) architectures require reach of several hundred km and the ability to pass through ten or more ROADMs. However, a number of fundamental challenges remain including the selection of appropriate modulation formats that are robust to a variety of nonlinearities, are sufficiently spectrally efficient, and able to withstand the strong optical filtering of cascaded ROADMs. Here we compare a variety of single-carrier quaternary modulation formats, each providing 2 bits/symbol/polarization and each likely to provide some advantages at 100Gb/s. Each format is presented with an appropriate MZM-based transmitter, and constrained by practical signal fidelity limitations that also enable comparison to experimental results from our 100G testbed. We primarily examine direct detection for cost-sensitive metro networks; however we also quantify the performance of coherent receivers, where applicable. Simulation results demonstrate the relative OSNR penalty (at a pre-FEC BER of 10-3) for a range of launch powers and adjacent channel formats.